Horizontal and endosymbiotic gene transfer in early plastid evolution

Summary

Plastids evolved from a cyanobacterium that was engulfed by a heterotrophic eukaryotic host and became a stable organelle. Some of the resulting eukaryotic algae entered into a number of secondary endosymbioses with diverse eukaryotic hosts. These events had major consequences on the evolution and diversification of life on Earth. Although almost all plastid diversity derives from a single endosymbiotic event, analysis of nuclear genomes of plastid-bearing lineages has revealed a mosaic origin of plastid-related genes. In addition to cyanobacterial genes, plastids recruited for their functioning eukaryotic proteins encoded by the host nucleus and also bacterial proteins of non-cyanobacterial origin. Thus, plastid proteins and plastid-localized metabolic pathways evolved by tinkering using the gene toolkits provided by the cyanobacterial endosymbiont, its eukaryotic host, and other bacteria. This mixed heritage seems especially complex in secondary algae containing green plastids, the acquisition of which appears to have been facilitated by many previous acquisitions of red algal genes (the “red carpet hypothesis”).

I. Introduction

Oxygenic photosynthesis in eukaryotes appeared >1 billion years ago (Eme et al., 2014) via the endosymbiosis of a close relative of the deep-branching cyanobacterium Gloeomargarita lithophora within a phagotrophic eukaryotic host (Ponce-Toledo et al., 2017). Subsequently, the cyanobacterium evolved into a permanent photosynthetic organelle called primary plastid. This endosymbiosis is at the origin of the Archaeplastida, a monophyletic supergroup composed of three primary plastid-bearing lineages: the green algae plus land plants, the red algae, and the glaucophytes (Fig. 1; Adl et al., 2012). The cercozoan amoeba Paulinella chromatophora (Fig. 2) is the only known lineage apart from Archaeplastida in which an independent type of primary plastids evolved (Marin et al., 2005). The Paulinella plastids, called "chromatophores", derive from a cyanobacterium of the Synechococcus/Prochlorococcus group, in contrast with the Gloeomargarita-like plastids of Archaeplastida. Whereas primary endosymbioses are extremely rare in biological history, with only the Archaeplastida and Paulinella cases known, green and red algae have participated as endosymbionts in numerous secondary and tertiary endosymbioses originating a broad diversity of eukaryotic lineages with complex plastids (Fig. 1; Moreira & Philippe, 2001; Keeling, 2010).

Fig. 1.

The distribution of photosynthesis in the eukaryotic global phylogeny. Colored solid branches correspond to photosynthetic lineages endowed with primary plastids and colored dashed branches to lineages with secondary plastids (green and red colors indicate the type of secondary endosymbiont, green or red algae, respectively). Blue arrows show the two known primary endosymbioses (in Archaeplastida and Paulinella) and green and red arrows indicate the secondary endosymbioses involving green and red algal endosymbionts. Grey branches correspond to non-photosynthetic eukaryotic phyla. The tree has been largely modified from Adl et al., 2012.

Fig. 2.

Light microscopy image of Paulinella chromatophora. The pigmented Synechococcus-like primary plastids are easily visible within the cytoplasm. Scale bar: 10 µm. Image courtesy of Eva Nowack (Heinrich-Heine-Universität Düsseldorf).

The reduced number of primary plastid acquisitions compared to that of eukaryotic lineages with complex plastids suggests that the enslavement of a photosynthetic cyanobacterial endosymbiont (or cyanobiont) is, for unknown reasons, more challenging from an evolutionary point of view and requires specific adaptations to stabilize the cyanobiont as a permanent organelle. Once the primary plastid was fully integrated within the host (through the evolution of metabolite export and protein import systems, transfer of plastid genes to the host nucleus, and evolution of proteins involved in redox regulation) it was easier for algae to become endosymbionts of other eukaryotes. Here, we review the genetic and genomic changes that accompanied the evolution of primary plastids and explore the plastid proteome composition to propose a possible role for host-derived and exogenous genes in the establishment of these plastids. Likewise, we discuss the genetic mosaicism of nuclear genomes in complex plastid-harboring lineages in the light of horizontal gene transfer (HGT) and putative cryptic endosymbioses.

II. Evolution of primary plastids in Archaeplastida

Plastids in Archaeplastida derive from an endosymbiotic cyanobacterium that was fully integrated in a heterotrophic eukaryotic host and became an organelle. Nonetheless, plastids differ considerably from free-living cyanobacteria. One of the most remarkable differences is the drastic reduction of plastid genomes, which encode less than 5% of the genes found in typical free-living cyanobacteria (Green, 2011). The number of protein-coding genes in plastid genomes varies across Archaeplastida, with red algae harboring the largest sets (160-235 proteins) (Lee et al., 2016), perhaps as a result of the massive gene loss in the nuclear genome likely experienced by the rhodophyte ancestor, which exerted a selective pressure to keep plastid-encoded genes (Bhattacharya et al., 2018).

Gene loss in the cyanobiont was a crucial event during early endosymbiosis stages and marked the transformation of the cyanobacterium into an obligatory endosymbiont. Glycogen synthase genes, absent in plastids, were possibly among the first genes to be lost and this may have helped to achieve the early host-endosymbiont metabolic integration (Gavelis & Gile, 2018). Glycogen synthase defective mutant cyanobacteria release organic carbon outside the cell as a homeostatic response to cope with excessive photosynthate production and the inability to store it as glycogen (Cano et al., 2018). Therefore, it is possible that inefficient glycogen storage allowed the leakage of photosynthate that could be used by the host before the evolution of a specific and more efficient photosynthate export system.

It is commonly assumed that endosymbiotic gene transfer (EGT) was critical for plastid endosymbiosis because it allowed the host to gain control over the expression of plastid genes, thus increasing the dependence of the cyanobiont upon the host gene expression machinery. The majority of plastid genes transferred to the host nucleus participates in photosynthesis-related functions (e.g., photosystem subunits, chlorophyll synthesis) and plastid metabolism and maintenance (e.g., amino acid synthesis and plastid division) (Reyes-Prieto et al., 2006). Recent studies have suggested that EGTs contributed also to expand the redox sensing capabilities of the host, sometimes through genetic tinkering that created new chimeric proteins (Méheust et al., 2016). This process was accompanied by the evolution of new redox sensing pathways that helped the host to cope with the increased reactive oxygen species concentration produced by the new organelle (Woehle et al., 2017).

Even though plastid-encoded proteins plus plastid-targeted EGT products are essential for the correct functioning of these organelles, they represent less than half of the total plastid proteome in Archaeplastida (Qiu et al., 2013). The fact that the plastid proteome contains a vast proportion of apparently non-cyanobacterial proteins opens interesting questions about the mechanisms underlying plastid acquisition and suggest that plastid symbiogenesis was not as straightforward as it is commonly assumed. For instance, recent phylogenetic analyses suggest that the genetic machinery necessary to synthesize the galactolipids of plastid membranes -once thought to derive from the cyanobiont- do not have cyanobacterial but eukaryotic origin, opening many questions about the nature of plastid membranes (Sato & Awai, 2017).

Host-derived proteins, evolved either from the retargeting of pre-existing proteins or from new gene innovations, participate in a wide range of plastid functions and represent the largest fraction of plastid-targeted proteins (Qiu et al., 2013). For instance, metabolite transporters originated from the retargeting of host membrane proteins are particularly overrepresented in the plastid envelope (58% of plastid transporters appear to derive from host membrane proteins in Arabidopsis thaliana; Tyra et al., 2007). These findings suggest that the host drove the early integration of the cyanobiont by providing the proteins necessary to connect the cyanobiont metabolism with the energy demands of the host (“host-centric” endosymbiosis model) (Karkar et al., 2015).

In addition to host-derived genes, non-cyanobacterial bacterial genes were also crucial for plastid evolution and represent 7-15% of the plastid proteome (Qiu et al., 2013). The phylogenetic origin of these non-cyanobacterial prokaryotic genes in Archaeplastida seems to cover a wide range of bacterial phyla (Dagan et al., 2013). After Cyanobacteria, Proteobacteria appear to be the most common contributors to the plastid proteome, particularly alphaproteobacterial proteins that may derive from the mitochondrial ancestor (Dagan et al., 2013; Qiu et al., 2013). Many of the bacterial genes were likely acquired horizontally due to the presumed phagotrophic lifestyle of the Archaeplastida ancestor and may have helped to compensate the massive gene loss undergone by the cyanobiont genome.

An issue that has attracted much attention is the presence of a number of genes apparently transferred from Chlamydiales bacteria to the Archaeplastida (Huang & Gogarten, 2007; Becker et al., 2008), which has even led to propose a tripartite model of plastid origin (Facchinelli et al., 2013). This model suggests that chlamydial cells infected the Archaeplastida ancestor and that this infection helped in the early steps of plastid acquisition by protecting the cyanobiont from host defenses and supplying multiple enzymes to integrate the photosynthate produced by the cyanobiont into the host carbohydrate metabolism (Facchinelli et al., 2013) and/or by allowing the cyanobiont to cope with ATP starvation as a result of the hypoxic environment of the host cytosol (Cenci et al., 2018). However, this model encounters several problems. First, there is no report of any Chlamydia species able to infect Archaeplastida, which suggests that the Archaeplastida ancestor was not a likely host for this family of pathogenic bacteria. Second, and more importantly, phylogenetic reanalyses of putative Chlamydia-derived genes have detected various phylogenetic artefacts and reduced considerably the number of genes compatible with a putative chlamydial ancestry (Deschamps, 2014, Moreira & Deschamps, 2014). Hence, the tripartite model of plastid origin continues to be the subject of hot debates.

III. Chromatophore evolution in Paulinella chromatophora

The cercozoan amoeba Paulinella chromatophora (Fig. 2) adopted a primary plastid (called chromatophore) by the endosymbiosis of a cyanobacterium closely related to the Synechococcus/Prochlorococcus (Syn/Pro) clade (Marin et al., 2005). Paulinella is a good model to study the early evolution of primary plastids since the divergence of the chromatophore from its Syn/Pro ancestor is relatively recent, only 90-140 Mya (Delaye et al., 2016). Remarkably, there are important similarities between the primary endosymbioses in Archaeplastida and P. chromatophora, likely due to convergent evolution in the process of plastid acquisition (Table 1).

Table 1. Comparison of plastid characters between Archaeplastida and Paulinella chromatophora.

Character	Archaeplastida	Paulinella	Reference
Plastid genome size	100-200 kbp	1021 kbp	Nowack et al., 2008
Number of plastid genes	80-250	911	Nowack et al., 2008
EGTs in plastid proteome	70-390	>70	Qiu et al., 2013; Singer et al., 2017; Nowack et al., 2016; Zhang et al., 2017
Non-cyanobacterial prokaryotic proteins in plastid proteome	40-240	>170	Qiu et al., 2013; Singer et al., 2017; Nowack et al., 2016; Zhang et al., 2017
Host-derived and proteins of uncertain origin in plastid proteome	320-900	>390	Qiu et al., 2013; Singer et al., 2017; Nowack et al., 2016; Zhang et al., 2017
Lineage age estimation	>1000 Myr	90-140 Myr	Eme et al., 2014; Delaye et al., 2016
Import system of nucleus-encoded proteins into the plastid	Translocons at the outer and inner plastid membranes (TOC/TIC complex)	Vesicles of the host endomembrane system fusion with the outer plastid membrane and proteins cross the inner membrane through a simplified TIC translocon	Mackiewicz et al., 2012
Phagotrophic capacity	Lost in Rhodophyta, Glaucophyta, and most Viridiplantae but preserved in some prasinophytes	Lost	Gagat & Mackiewicz, 2017
Peptidoglycan wall	Present in Glaucophyta and in some Viridiplantae species	Present	Gagat & Mackiewicz, 2017

The chromatophore genome is highly reduced (encoding 867 proteins, which represent about 1/3 of proteins of its free-living counterparts (Nowack et al., 2008). Similar to the EGTs found in Archaeplastida, P. chromatophora has relocated more than 70 chromatophore genes into the nuclear genome (mostly involved in photosynthesis-related functions) (Nowack et al., 2011, 2016; Zhang et al., 2017). By contrast, these genes represent less than 1% of the Paulinella nuclear genome, while in A. thaliana some reports suggest that the genes of cyanobacterial origin can account up to 18% of the nuclear genes (Martin et al., 2002). Nonetheless, the chromatophore genome reduction is most likely still ongoing and it is possible that more genes will be transferred to the host nucleus. At any rate, based on the reduced number of chromatophore-derived genes in the nuclear genome, it seems that EGT may have been less important in the establishment of the chromatophore in P. chromatophora than HGT from other bacteria, as about 170 genes of bacterial origin encode proteins likely targeted to the chromatophore (Nowack et al., 2016). Interestingly, the largest contribution of identified chromatophore-targeted proteins derive from the ancestral host genetic repertoire (Singer et al., 2017), a similar pattern as the one observed in Archaeplastida proteomes (Qiu et al., 2013), suggesting that in both primary endosymbioses the host played a crucial role to mediate the cyanobiont integration.

IV. Complex plastids

Eukaryotic lineages with complex plastids evolved by the engulfment of red or green algae by different heterotrophic hosts. This type of event, called secondary endosymbiosis, originated plastids with three or four membranes (in contrast with the two membranes of primary plastids) and introduced a high degree of reticulation within the eukaryotic global phylogeny that is not yet fully understood (Archibald, 2015). Four ecologically diverse eukaryotic lineages have red alga-derived plastids: cryptophytes, alveolates, stramenopiles, and haptophytes (the “CASH” assemblage). While the phylogeny of plastid genes supports that a single red alga is at the origin of all complex red plastids (Yoon et al., 2002), most phylogenetic analyses of nuclear genes suggest that CASH lineages are not monophyletic, which has been interpreted as an indication that complex red plastids were acquired through an undetermined number of serial endosymbioses with different hosts (Bodył et al., 2009; Baurain et al., 2010; Petersen et al., 2014). Conversely, the origin of complex green plastids is much clearer: euglenids and chlorarachniophytes acquired their plastids from two independent secondary endosymbioses involving two distantly related green algal endosymbionts (Jackson et al., 2018).

Similar to primary endosymbiosis, during secondary endosymbiotic events genes were transferred from the algal endosymbiont to the host nucleus (secondary EGTs). The analyses of nuclear genes with algal affiliation in eukaryotic lineages with complex plastids have revealed a wide range of putative algal donors different from the algal endosymbionts that exist today as secondary plastids (Curtis et al., 2012). For instance, despite diatoms contain plastids clearly derived from red algae, phylogenetic analysis of diatom nuclear genomes suggested that more than 1,700 nucleus-encoded genes were apparently transferred from green algae (Moustafa et al., 2009). Likewise, recent estimations suggest that ~25% of nucleus-encoded plastid-targeted proteins in the ancestor of ochrophytes (photosynthetic stramenopiles) derive from green algae (Dorrell et al., 2017). This apparent massive genetic mosaicism may be explained either as the result of high frequency of eukaryote-to-eukaryote HGT or as the consequence of putative cryptic endosymbioses. Nevertheless, although it is possible that the “green” signal observed in ochrophyte genomes, particularly in diatoms, might attest for a former endosymbiosis with a green alga (see Moustafa et al., 2009), it seems more likely that the green contribution to diatom genomes has been largely overestimated because of several undetected tree reconstruction artefacts (Deschamps & Moreira, 2012).

V. The “red carpet” hypothesis

For some secondary EGTs, phylogenetic analysis allows to trace back the full sequence of endosymbiotic gene transfer from cyanobacteria to the Archaeplastida ancestor and then from red or green algal endosymbionts to the secondary photosynthetic lineages. Phylogenetic inspection of these EGTs in the CASH lineages has shown that -as expected- the majority of these genes were transferred from the red algal secondary endosymbiont (Deschamps & Moreira, 2012; Ponce-Toledo et al., 2018). By contrast, >30% and >50% of this this type of cyanobacteria-derived secondary EGTs appear also to have red algal ancestry in euglenids and chlorarachniophytes, respectively (Ponce-Toledo et al., 2018). Thus, these two green plastid-harboring lineages have an unexpected mix of red and green plastid-targeted proteins that creates a high mosaicism in their plastid metabolic pathways (e.g. biosynthetic pathways, photosynthesis-related functions, plastid biogenesis) (Yang et al., 2011). Interestingly, the dinoflagellate Lepidodinium chlorophorum, which is a clear case of replacement of an original red algal endosymbiont by a green algal one (Saldarriaga et al., 2001), harbors several nucleus-encoded genes transferred from the former red plastid that were retargeted to the new green plastid. Thus, the current green algal plastid functions with a mix of red and green genes (Minge et al., 2010).

However, in the case of euglenids and chlorarachniophytes, it is difficult to know if the presence of this large amount of genes of red algal origin is due to the fact that these lineages also experienced former endosymbioses with red algae. Cryptic endosymbiosis scenarios have to be considered cautiously when interpreting the chimerism observed in the nuclear genomes of photosynthetic eukaryotes (Deschamps & Moreira, 2012). Molecular clock analyses suggest that chlorarachniophytes acquired their green plastid 578-318 Mya (Jackson et al., 2018) while the chlorarachniophyte host lineage appears to have diverged from heterotrophic cercozoa ~1000 Mya (Parfrey et al., 2011). Therefore, there was a long period (>400 million years) during which a putative secondary endosymbiosis with a red alga might have taken place before the acquisition of the present green algal plastid. What is certain is that red algae (or lineages containing red algal secondary plastids) provided many plastid-related genes to both euglenids and chlorarachniophytes. These genes appear to be present in all known species of each of these two phyla, indicating that they correspond to ancient gene acquisitions before the diversification of these two phyla (Ponce-Toledo et al., 2018). We introduce here the “red carpet hypothesis” to propose that these red algal genes, transferred to the host nucleus before and/or during the early steps of endosymbiosis with green algae, provided important plastid-related functions and acted as a sort of “red carpet” to facilitate the subsequent adoption of the new green algal endosymbionts.

VI. Conclusions

Phylogenetic analyses of nucleus-encoded plastid-targeted proteins have revealed the massive contribution of non-cyanobacterial proteins to the plastid proteomes of Archaeplastida and P. chromatophora, the only two lineages known to harbor primary plastids (Qiu et al., 2013; Nowack et al., 2016). Gene transfers from bacteria other than the cyanobiont and their retargeting to the early plastid seem to have been very frequent in both primary photosynthetic lineages, many possibly replacing genes that were lost in the cyanobiont genome. Nonetheless, the eukaryotic host seems to have been the largest contributor of plastid-targeted proteins (most of them involved in plastid maintenance and transport of metabolites), which supports the idea that the host drove the early steps of plastid endosymbiosis (Karkar et al., 2015).

It is still not completely clear why secondary endosymbioses have been much more recurrent than primary ones in the evolution of contemporary eukaryotic photosynthetic lineages. However, it seems that nucleus-encoded genes acquired from previous endosymbioses can be helpful to regain a plastid, likely because the expression of these genes is already under the control of the host and because they carry targeting signals that can be more easily reused to target proteins towards a new endosymbiont (Matsuo & Inagaki, 2018), thus speeding up the host-endosymbiont integration. Although cryptic endosymbioses are difficult to prove, we postulate that plastid-targeted secondary EGTs are critical markers to test these evolutionary scenarios.